Children's early vocabularies of their first 50 to 100 words provide one line of evidence suggesting that object categories are more easily learned than relational categories (see Gentner, gentner:82; Gentner & Boroditsky gentner+boroditsky for reviews and discussion). Indeed, for a long time, object names were considered privileged at the start of word learning. Certainly common nouns dominate in English-speaking children's early vocabularies; relational terms (verbs and spatial terms) are rare [GentnerGentner1982,MacnamaraMacnamara1982,NelsonNelson1973]. The universality of the noun advantage across languages is currently under attack [Bloom, Tinker MargulisBloom 1993,TardifTardif1996,Gopnik ChoiGopnik Choi1995]. In particular, it has been suggested that in certain languages such as Korean [Gopnik ChoiGopnik Choi1995] and Mandarin [TardifTardif1996] verbs are as or more prevalent in early vocabularies. There are many controversies in this literature and studies contradicting the claim of a verb advantage in Korean and Mandarin [Au, Dapretto SongAu 1994,GentnerGentner1982,TardifTardif1996]. The controversies and ambiguities all revolve around how to measure the words in children's vocabularies. However, the best summary at present is that when proper nouns are included in the counts, the noun advantage over relational terms varies from a 4:1 noun advantage in early vocabulary by some measures in some languages [Au, Dapretto SongAu 1994] to near 1:1 parity by other measures in other languages (Tardif, 1996; see Gentner & Boroditsky, in press for a more detailed discussion.) All in all, there would seem to be a bias toward learning nouns over relational terms in early word learning, but just as clearly this advantage depends to some degree on the language being learned and on parents' usual means of talking to their children.
The evidence on early vocabularies concerns the words children say without consideration of how they understand them. Stronger evidence for the claim that object concepts are easier and earlier than relational concepts derives not from children's word productions but from their comprehension of common nouns versus relational terms. There are extensive literatures in both areas although they are difficult to compare because of differing methods, ages of subjects, and empirical questions. These differences derive directly from the apparent ease with which children learn object categories as opposed to their difficulty in learning relational categories. The key question for researchers who study early noun acquisition is how it is that children learn so many nouns so rapidly and with so few errors. The only errors consistently studied in this literature are the overextension errors in production typically noticed at about the time productive vocabulary growth begins to accelerate. However, these errors may not be category errors per se. Instead, these overextensions (for example, calling a zebra ``doggie'') may reflect pragmatic strategies or retrieval errors [Gershkoff-Stowe SmithGershkoff-Stowe Smith1997,HuttenlocherHuttenlocher1974]. Consistent with this idea is the rarity of overextensions in comprehension (see, for example, Naigles & Gelman, 1995); if production overextensions result from overly broad representations of noun meaning categories, those representations should enter into the comprehension of those nouns as well.
In contrast, the key question for researchers who study the acquisition of relational terms is why they are so difficult to learn. The central phenomena are comprehension errors. Long after children begin to use relational terms, when they are as old as three, four, or even five years, their interpretations are errorful. Preschool children make such mistakes as misinterpreting in front of to mean `near,' put to mean `give,' higher to mean `on top' [Johnston SlobinJohnston Slobin1979,BowermanBowerman1994,ClarkClark1971,GentnerGentner1975,Kuczaj MaratsosKuczaj Maratsos1975,Smith, Cooney McCordSmith 1986]). Simply, common object categories are for the most part trivially easy for children to acquire whereas relational terms exhibit a protracted and errorful course of development.
This difference is also evident in artificial word learning studies. In these studies researchers present children with a novel object or event and label it. Children's interpretation of the label is measured by the kinds of other objects to which they generalize the newly learned label. Considerable evidence indicates that by 18 months (and quite possibly before), children systematically generalize novel nouns to new instances that are in the same taxonomic category [MarkmanMarkman1989,WaxmanWaxman1994,SmithSmith1995]. There are fewer studies of children's generalizations of novel relational terms and most involve children at least 3 years of age and older. But the results of these studies are markedly different from those concerning object terms. First, young children's judgments are more variable and less systematic [LandauLandau1996]; they sometimes err by interpreting relational terms as labels for one of the objects entering in the relation [Kersten SmithKersten Smith1998,Ryalls, Winslow SmithRyalls in press]; and they are much more conservative in their generalizations [Tomasello, Akhtar, Dodson RekauTomasello 1997].
In sum, one major fact to be explained in developing a theory of relational representations is the relative difficulty in acquiring words to talk about relations as opposed to words to talk about objects.